]> git.saurik.com Git - wxWidgets.git/blob - src/richtext/richtexthtml.cpp
Fixed tab removal bug, and empty text element
[wxWidgets.git] / src / richtext / richtexthtml.cpp
1 /////////////////////////////////////////////////////////////////////////////
2 // Name: src/richtext/richtexthtml.cpp
3 // Purpose: HTML I/O for wxRichTextCtrl
4 // Author: Julian Smart
5 // Modified by:
6 // Created: 2005-09-30
7 // RCS-ID: $Id$
8 // Copyright: (c) Julian Smart
9 // Licence: wxWindows licence
10 /////////////////////////////////////////////////////////////////////////////
11
12 // For compilers that support precompilation, includes "wx.h".
13 #include "wx/wxprec.h"
14
15 #ifdef __BORLANDC__
16 #pragma hdrstop
17 #endif
18
19 #if wxUSE_RICHTEXT
20
21 #include "wx/richtext/richtexthtml.h"
22 #include "wx/richtext/richtextstyles.h"
23
24 #ifndef WX_PRECOMP
25 #endif
26
27 #include "wx/filename.h"
28 #include "wx/wfstream.h"
29 #include "wx/txtstrm.h"
30
31 #if wxUSE_FILESYSTEM
32 #include "wx/filesys.h"
33 #include "wx/fs_mem.h"
34 #endif
35
36 IMPLEMENT_DYNAMIC_CLASS(wxRichTextHTMLHandler, wxRichTextFileHandler)
37
38 int wxRichTextHTMLHandler::sm_fileCounter = 1;
39
40 wxRichTextHTMLHandler::wxRichTextHTMLHandler(const wxString& name, const wxString& ext, int type)
41 : wxRichTextFileHandler(name, ext, type), m_buffer(NULL), m_font(false), m_inTable(false)
42 {
43 m_fontSizeMapping.Add(8);
44 m_fontSizeMapping.Add(10);
45 m_fontSizeMapping.Add(13);
46 m_fontSizeMapping.Add(17);
47 m_fontSizeMapping.Add(22);
48 m_fontSizeMapping.Add(30);
49 m_fontSizeMapping.Add(100);
50 }
51
52 /// Can we handle this filename (if using files)? By default, checks the extension.
53 bool wxRichTextHTMLHandler::CanHandle(const wxString& filename) const
54 {
55 wxString path, file, ext;
56 wxFileName::SplitPath(filename, & path, & file, & ext);
57
58 return (ext.Lower() == wxT("html") || ext.Lower() == wxT("htm"));
59 }
60
61
62 #if wxUSE_STREAMS
63 bool wxRichTextHTMLHandler::DoLoadFile(wxRichTextBuffer *WXUNUSED(buffer), wxInputStream& WXUNUSED(stream))
64 {
65 return false;
66 }
67
68 /*
69 * We need to output only _changes_ in character formatting.
70 */
71
72 bool wxRichTextHTMLHandler::DoSaveFile(wxRichTextBuffer *buffer, wxOutputStream& stream)
73 {
74 m_buffer = buffer;
75
76 ClearTemporaryImageLocations();
77
78 buffer->Defragment();
79
80 #if wxUSE_UNICODE
81 wxCSConv* customEncoding = NULL;
82 wxMBConv* conv = NULL;
83 if (!GetEncoding().IsEmpty())
84 {
85 customEncoding = new wxCSConv(GetEncoding());
86 if (!customEncoding->IsOk())
87 {
88 delete customEncoding;
89 customEncoding = NULL;
90 }
91 }
92 if (customEncoding)
93 conv = customEncoding;
94 else
95 conv = & wxConvUTF8;
96 #endif
97
98 {
99 #if wxUSE_UNICODE
100 wxTextOutputStream str(stream, wxEOL_NATIVE, *conv);
101 #else
102 wxTextOutputStream str(stream, wxEOL_NATIVE);
103 #endif
104
105 wxTextAttr currentParaStyle = buffer->GetAttributes();
106 wxTextAttr currentCharStyle = buffer->GetAttributes();
107
108 if ((GetFlags() & wxRICHTEXT_HANDLER_NO_HEADER_FOOTER) == 0)
109 str << wxT("<html><head></head><body>\n");
110
111 OutputFont(currentParaStyle, str);
112
113 m_font = false;
114 m_inTable = false;
115
116 m_indents.Clear();
117 m_listTypes.Clear();
118
119 wxRichTextObjectList::compatibility_iterator node = buffer->GetChildren().GetFirst();
120 while (node)
121 {
122 wxRichTextParagraph* para = wxDynamicCast(node->GetData(), wxRichTextParagraph);
123 wxASSERT (para != NULL);
124
125 if (para)
126 {
127 wxTextAttr paraStyle(para->GetCombinedAttributes());
128
129 BeginParagraphFormatting(currentParaStyle, paraStyle, str);
130
131 wxRichTextObjectList::compatibility_iterator node2 = para->GetChildren().GetFirst();
132 while (node2)
133 {
134 wxRichTextObject* obj = node2->GetData();
135 wxRichTextPlainText* textObj = wxDynamicCast(obj, wxRichTextPlainText);
136 if (textObj && !textObj->IsEmpty())
137 {
138 wxTextAttr charStyle(para->GetCombinedAttributes(obj->GetAttributes()));
139 BeginCharacterFormatting(currentCharStyle, charStyle, paraStyle, str);
140
141 wxString text = textObj->GetText();
142
143 if (charStyle.HasTextEffects() && (charStyle.GetTextEffects() & wxTEXT_ATTR_EFFECT_CAPITALS))
144 text.MakeUpper();
145
146 wxString toReplace = wxRichTextLineBreakChar;
147 text.Replace(toReplace, wxT("<br>"));
148
149 str << text;
150
151 EndCharacterFormatting(currentCharStyle, charStyle, paraStyle, str);
152 }
153
154 wxRichTextImage* image = wxDynamicCast(obj, wxRichTextImage);
155 if( image && (!image->IsEmpty() || image->GetImageBlock().GetData()))
156 WriteImage( image, stream );
157
158 node2 = node2->GetNext();
159 }
160
161 EndParagraphFormatting(currentParaStyle, paraStyle, str);
162
163 str << wxT("\n");
164 }
165 node = node->GetNext();
166 }
167
168 CloseLists(-1, str);
169
170 str << wxT("</font>");
171
172 if ((GetFlags() & wxRICHTEXT_HANDLER_NO_HEADER_FOOTER) == 0)
173 str << wxT("</body></html>");
174
175 str << wxT("\n");
176 }
177
178 #if wxUSE_UNICODE
179 if (customEncoding)
180 delete customEncoding;
181 #endif
182
183 m_buffer = NULL;
184
185 return true;
186 }
187
188 void wxRichTextHTMLHandler::BeginCharacterFormatting(const wxTextAttr& currentStyle, const wxTextAttr& thisStyle, const wxTextAttr& WXUNUSED(paraStyle), wxTextOutputStream& str)
189 {
190 wxString style;
191
192 // Is there any change in the font properties of the item?
193 if (thisStyle.GetFontFaceName() != currentStyle.GetFontFaceName())
194 {
195 wxString faceName(thisStyle.GetFontFaceName());
196 style += wxString::Format(wxT(" face=\"%s\""), faceName.c_str());
197 }
198 if (thisStyle.GetFontSize() != currentStyle.GetFontSize())
199 style += wxString::Format(wxT(" size=\"%ld\""), PtToSize(thisStyle.GetFontSize()));
200 if (thisStyle.GetTextColour() != currentStyle.GetTextColour() )
201 {
202 wxString color(thisStyle.GetTextColour().GetAsString(wxC2S_HTML_SYNTAX));
203 style += wxString::Format(wxT(" color=\"%s\""), color.c_str());
204 }
205
206 if (style.size())
207 {
208 str << wxString::Format(wxT("<font %s >"), style.c_str());
209 m_font = true;
210 }
211
212 if (thisStyle.GetFontWeight() == wxBOLD)
213 str << wxT("<b>");
214 if (thisStyle.GetFontStyle() == wxITALIC)
215 str << wxT("<i>");
216 if (thisStyle.GetFontUnderlined())
217 str << wxT("<u>");
218
219 if (thisStyle.HasURL())
220 str << wxT("<a href=\"") << thisStyle.GetURL() << wxT("\">");
221 }
222
223 void wxRichTextHTMLHandler::EndCharacterFormatting(const wxTextAttr& WXUNUSED(currentStyle), const wxTextAttr& thisStyle, const wxTextAttr& WXUNUSED(paraStyle), wxTextOutputStream& stream)
224 {
225 if (thisStyle.HasURL())
226 stream << wxT("</a>");
227
228 if (thisStyle.GetFontUnderlined())
229 stream << wxT("</u>");
230 if (thisStyle.GetFontStyle() == wxITALIC)
231 stream << wxT("</i>");
232 if (thisStyle.GetFontWeight() == wxBOLD)
233 stream << wxT("</b>");
234
235 if (m_font)
236 {
237 m_font = false;
238 stream << wxT("</font>");
239 }
240 }
241
242 /// Begin paragraph formatting
243 void wxRichTextHTMLHandler::BeginParagraphFormatting(const wxTextAttr& WXUNUSED(currentStyle), const wxTextAttr& thisStyle, wxTextOutputStream& str)
244 {
245 if (thisStyle.HasPageBreak())
246 {
247 str << wxT("<div style=\"page-break-after:always\"></div>\n");
248 }
249
250 if (thisStyle.HasLeftIndent() && thisStyle.GetLeftIndent() != 0)
251 {
252 if (thisStyle.HasBulletStyle())
253 {
254 int indent = thisStyle.GetLeftIndent();
255
256 // Close levels high than this
257 CloseLists(indent, str);
258
259 if (m_indents.GetCount() > 0 && indent == m_indents.Last())
260 {
261 // Same level, no need to start a new list
262 }
263 else if (m_indents.GetCount() == 0 || indent > m_indents.Last())
264 {
265 m_indents.Add(indent);
266
267 wxString tag;
268 int listType = TypeOfList(thisStyle, tag);
269 m_listTypes.Add(listType);
270
271 // wxHTML needs an extra <p> before a list when using <p> ... </p> in previous paragraphs.
272 // TODO: pass a flag that indicates we're using wxHTML.
273 str << wxT("<p>\n");
274
275 str << tag;
276 }
277
278 str << wxT("<li> ");
279 }
280 else
281 {
282 CloseLists(-1, str);
283
284 wxString align = GetAlignment(thisStyle);
285 str << wxString::Format(wxT("<p align=\"%s\""), align.c_str());
286
287 wxString styleStr;
288
289 if ((GetFlags() & wxRICHTEXT_HANDLER_USE_CSS) && thisStyle.HasParagraphSpacingBefore())
290 {
291 float spacingBeforeMM = thisStyle.GetParagraphSpacingBefore() / 10.0;
292
293 styleStr += wxString::Format(wxT("margin-top: %.2fmm; "), spacingBeforeMM);
294 }
295 if ((GetFlags() & wxRICHTEXT_HANDLER_USE_CSS) && thisStyle.HasParagraphSpacingAfter())
296 {
297 float spacingAfterMM = thisStyle.GetParagraphSpacingAfter() / 10.0;
298
299 styleStr += wxString::Format(wxT("margin-bottom: %.2fmm; "), spacingAfterMM);
300 }
301
302 float indentLeftMM = (thisStyle.GetLeftIndent() + thisStyle.GetLeftSubIndent())/10.0;
303 if ((GetFlags() & wxRICHTEXT_HANDLER_USE_CSS) && (indentLeftMM > 0.0))
304 {
305 styleStr += wxString::Format(wxT("margin-left: %.2fmm; "), indentLeftMM);
306 }
307 float indentRightMM = thisStyle.GetRightIndent()/10.0;
308 if ((GetFlags() & wxRICHTEXT_HANDLER_USE_CSS) && thisStyle.HasRightIndent() && (indentRightMM > 0.0))
309 {
310 styleStr += wxString::Format(wxT("margin-right: %.2fmm; "), indentRightMM);
311 }
312 // First line indentation
313 float firstLineIndentMM = - thisStyle.GetLeftSubIndent() / 10.0;
314 if ((GetFlags() & wxRICHTEXT_HANDLER_USE_CSS) && (firstLineIndentMM > 0.0))
315 {
316 styleStr += wxString::Format(wxT("text-indent: %.2fmm; "), firstLineIndentMM);
317 }
318
319 if (!styleStr.IsEmpty())
320 str << wxT(" style=\"") << styleStr << wxT("\"");
321
322 str << wxT(">");
323
324 // TODO: convert to pixels
325 int indentPixels = indentLeftMM*10/4;
326
327 if ((GetFlags() & wxRICHTEXT_HANDLER_USE_CSS) == 0)
328 {
329 // Use a table to do indenting if we don't have CSS
330 str << wxString::Format(wxT("<table border=0 cellpadding=0 cellspacing=0><tr><td width=\"%d\"></td><td>"), indentPixels);
331 m_inTable = true;
332 }
333
334 OutputFont(thisStyle, str);
335
336 if (((GetFlags() & wxRICHTEXT_HANDLER_USE_CSS) == 0) && (thisStyle.GetLeftSubIndent() < 0))
337 {
338 str << SymbolicIndent( - thisStyle.GetLeftSubIndent());
339 }
340 }
341 }
342 else
343 {
344 CloseLists(-1, str);
345
346 wxString align = GetAlignment(thisStyle);
347 str << wxString::Format(wxT("<p align=\"%s\""), align.c_str());
348
349 wxString styleStr;
350
351 if ((GetFlags() & wxRICHTEXT_HANDLER_USE_CSS) && thisStyle.HasParagraphSpacingBefore())
352 {
353 float spacingBeforeMM = thisStyle.GetParagraphSpacingBefore() / 10.0;
354
355 styleStr += wxString::Format(wxT("margin-top: %.2fmm; "), spacingBeforeMM);
356 }
357 if ((GetFlags() & wxRICHTEXT_HANDLER_USE_CSS) && thisStyle.HasParagraphSpacingAfter())
358 {
359 float spacingAfterMM = thisStyle.GetParagraphSpacingAfter() / 10.0;
360
361 styleStr += wxString::Format(wxT("margin-bottom: %.2fmm; "), spacingAfterMM);
362 }
363
364 if (!styleStr.IsEmpty())
365 str << wxT(" style=\"") << styleStr << wxT("\"");
366
367 str << wxT(">");
368 }
369 }
370
371 /// End paragraph formatting
372 void wxRichTextHTMLHandler::EndParagraphFormatting(const wxTextAttr& WXUNUSED(currentStyle), const wxTextAttr& thisStyle, wxTextOutputStream& stream)
373 {
374 if (m_inTable)
375 {
376 if (thisStyle.HasFont())
377 stream << wxT("</font>");
378
379 stream << wxT("</td></tr></table></p>\n");
380 m_inTable = false;
381 }
382 else if (!thisStyle.HasBulletStyle())
383 stream << wxT("</p>\n");
384 }
385
386 /// Closes lists to level (-1 means close all)
387 void wxRichTextHTMLHandler::CloseLists(int level, wxTextOutputStream& str)
388 {
389 // Close levels high than this
390 int i = m_indents.GetCount()-1;
391 while (i >= 0)
392 {
393 int l = m_indents[i];
394 if (l > level)
395 {
396 if (m_listTypes[i] == 0)
397 str << wxT("</ol>");
398 else
399 str << wxT("</ul>");
400 m_indents.RemoveAt(i);
401 m_listTypes.RemoveAt(i);
402 }
403 else
404 break;
405 i --;
406 }
407 }
408
409 /// Output font tag
410 void wxRichTextHTMLHandler::OutputFont(const wxTextAttr& style, wxTextOutputStream& stream)
411 {
412 if (style.HasFont())
413 {
414 stream << wxString::Format(wxT("<font face=\"%s\" size=\"%ld\""), style.GetFontFaceName().c_str(), PtToSize(style.GetFontSize()));
415 if (style.HasTextColour())
416 stream << wxString::Format(wxT(" color=\"%s\""), style.GetTextColour().GetAsString(wxC2S_HTML_SYNTAX).c_str());
417 stream << wxT(" >");
418 }
419 }
420
421 int wxRichTextHTMLHandler::TypeOfList( const wxTextAttr& thisStyle, wxString& tag )
422 {
423 // We can use number attribute of li tag but not all the browsers support it.
424 // also wxHtmlWindow doesn't support type attribute.
425
426 bool m_is_ul = false;
427 if (thisStyle.GetBulletStyle() == (wxTEXT_ATTR_BULLET_STYLE_ARABIC|wxTEXT_ATTR_BULLET_STYLE_PERIOD))
428 tag = wxT("<ol type=\"1\">");
429 else if (thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_LETTERS_UPPER)
430 tag = wxT("<ol type=\"A\">");
431 else if (thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_LETTERS_LOWER)
432 tag = wxT("<ol type=\"a\">");
433 else if (thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_ROMAN_UPPER)
434 tag = wxT("<ol type=\"I\">");
435 else if (thisStyle.GetBulletStyle() == wxTEXT_ATTR_BULLET_STYLE_ROMAN_LOWER)
436 tag = wxT("<ol type=\"i\">");
437 else
438 {
439 tag = wxT("<ul>");
440 m_is_ul = true;
441 }
442
443 if (m_is_ul)
444 return 1;
445 else
446 return 0;
447 }
448
449 wxString wxRichTextHTMLHandler::GetAlignment( const wxTextAttr& thisStyle )
450 {
451 switch( thisStyle.GetAlignment() )
452 {
453 case wxTEXT_ALIGNMENT_LEFT:
454 return wxT("left");
455 case wxTEXT_ALIGNMENT_RIGHT:
456 return wxT("right");
457 case wxTEXT_ALIGNMENT_CENTER:
458 return wxT("center");
459 case wxTEXT_ALIGNMENT_JUSTIFIED:
460 return wxT("justify");
461 default:
462 return wxT("left");
463 }
464 }
465
466 void wxRichTextHTMLHandler::WriteImage(wxRichTextImage* image, wxOutputStream& stream)
467 {
468 wxTextOutputStream str(stream);
469
470 str << wxT("<img src=\"");
471
472 #if wxUSE_FILESYSTEM
473 if (GetFlags() & wxRICHTEXT_HANDLER_SAVE_IMAGES_TO_MEMORY)
474 {
475 if (!image->GetImage().Ok() && image->GetImageBlock().GetData())
476 image->LoadFromBlock();
477 if (image->GetImage().Ok() && !image->GetImageBlock().GetData())
478 image->MakeBlock();
479
480 if (image->GetImage().Ok())
481 {
482 wxString ext(image->GetImageBlock().GetExtension());
483 wxString tempFilename(wxString::Format(wxT("image%d.%s"), sm_fileCounter, ext));
484 wxMemoryFSHandler::AddFile(tempFilename, image->GetImage(), image->GetImageBlock().GetImageType());
485
486 m_imageLocations.Add(tempFilename);
487
488 str << wxT("memory:") << tempFilename;
489 }
490 else
491 str << wxT("memory:?");
492
493 sm_fileCounter ++;
494 }
495 else if (GetFlags() & wxRICHTEXT_HANDLER_SAVE_IMAGES_TO_FILES)
496 {
497 if (!image->GetImage().Ok() && image->GetImageBlock().GetData())
498 image->LoadFromBlock();
499 if (image->GetImage().Ok() && !image->GetImageBlock().GetData())
500 image->MakeBlock();
501
502 if (image->GetImage().Ok())
503 {
504 wxString tempDir(GetTempDir());
505 if (tempDir.IsEmpty())
506 tempDir = wxFileName::GetTempDir();
507
508 wxString ext(image->GetImageBlock().GetExtension());
509 wxString tempFilename(wxString::Format(wxT("%s/image%d.%s"), tempDir, sm_fileCounter, ext));
510 image->GetImageBlock().Write(tempFilename);
511
512 m_imageLocations.Add(tempFilename);
513
514 str << wxFileSystem::FileNameToURL(tempFilename);
515 }
516 else
517 str << wxT("file:?");
518
519 sm_fileCounter ++;
520 }
521 else // if (GetFlags() & wxRICHTEXT_HANDLER_SAVE_IMAGES_TO_BASE64) // this is implied
522 #endif
523 {
524 str << wxT("data:");
525 str << GetMimeType(image->GetImageBlock().GetImageType());
526 str << wxT(";base64,");
527
528 if (image->GetImage().Ok() && !image->GetImageBlock().GetData())
529 image->MakeBlock();
530
531 wxChar* data = b64enc( image->GetImageBlock().GetData(), image->GetImageBlock().GetDataSize() );
532 str << data;
533
534 delete[] data;
535 }
536
537 str << wxT("\" />");
538 }
539
540 long wxRichTextHTMLHandler::PtToSize(long size)
541 {
542 int i;
543 int len = m_fontSizeMapping.GetCount();
544 for (i = 0; i < len; i++)
545 if (size <= m_fontSizeMapping[i])
546 return i+1;
547 return 7;
548 }
549
550 wxString wxRichTextHTMLHandler::SymbolicIndent(long indent)
551 {
552 wxString in;
553 for(;indent > 0; indent -= 20)
554 in.Append( wxT("&nbsp;") );
555 return in;
556 }
557
558 const wxChar* wxRichTextHTMLHandler::GetMimeType(int imageType)
559 {
560 switch(imageType)
561 {
562 case wxBITMAP_TYPE_BMP:
563 return wxT("image/bmp");
564 case wxBITMAP_TYPE_TIF:
565 return wxT("image/tiff");
566 case wxBITMAP_TYPE_GIF:
567 return wxT("image/gif");
568 case wxBITMAP_TYPE_PNG:
569 return wxT("image/png");
570 case wxBITMAP_TYPE_JPEG:
571 return wxT("image/jpeg");
572 default:
573 return wxT("image/unknown");
574 }
575 }
576
577 // exim-style base64 encoder
578 wxChar* wxRichTextHTMLHandler::b64enc( unsigned char* input, size_t in_len )
579 {
580 // elements of enc64 array must be 8 bit values
581 // otherwise encoder will fail
582 // hmmm.. Does wxT macro define a char as 16 bit value
583 // when compiling with UNICODE option?
584 static const wxChar enc64[] = wxT("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/");
585 wxChar* output = new wxChar[4*((in_len+2)/3)+1];
586 wxChar* p = output;
587
588 while( in_len-- > 0 )
589 {
590 register wxChar a, b;
591
592 a = *input++;
593
594 *p++ = enc64[ (a >> 2) & 0x3f ];
595
596 if( in_len-- == 0 )
597 {
598 *p++ = enc64[ (a << 4 ) & 0x30 ];
599 *p++ = '=';
600 *p++ = '=';
601 break;
602 }
603
604 b = *input++;
605
606 *p++ = enc64[(( a << 4 ) | ((b >> 4) &0xf )) & 0x3f];
607
608 if( in_len-- == 0 )
609 {
610 *p++ = enc64[ (b << 2) & 0x3f ];
611 *p++ = '=';
612 break;
613 }
614
615 a = *input++;
616
617 *p++ = enc64[ ((( b << 2 ) & 0x3f ) | ((a >> 6)& 0x3)) & 0x3f ];
618
619 *p++ = enc64[ a & 0x3f ];
620 }
621 *p = 0;
622
623 return output;
624 }
625 #endif
626 // wxUSE_STREAMS
627
628 /// Delete the in-memory or temporary files generated by the last operation
629 bool wxRichTextHTMLHandler::DeleteTemporaryImages()
630 {
631 return DeleteTemporaryImages(GetFlags(), m_imageLocations);
632 }
633
634 /// Delete the in-memory or temporary files generated by the last operation
635 bool wxRichTextHTMLHandler::DeleteTemporaryImages(int flags, const wxArrayString& imageLocations)
636 {
637 size_t i;
638 for (i = 0; i < imageLocations.GetCount(); i++)
639 {
640 wxString location = imageLocations[i];
641
642 if (flags & wxRICHTEXT_HANDLER_SAVE_IMAGES_TO_MEMORY)
643 {
644 #if wxUSE_FILESYSTEM
645 wxMemoryFSHandler::RemoveFile(location);
646 #endif
647 }
648 else if (flags & wxRICHTEXT_HANDLER_SAVE_IMAGES_TO_FILES)
649 {
650 if (wxFileExists(location))
651 wxRemoveFile(location);
652 }
653 }
654
655 return true;
656 }
657
658
659 #endif
660 // wxUSE_RICHTEXT
661